824 BELL SYSTEM TECHNICAL JOURNAL 



Several procedures are followed in order to correct, in so far as 

 possible, for practice effects. In comparative tests, whenever it is 

 possible, the circuits to be compared are interspersed so as to average 

 out practice effects. If it is desired to compare the articulation of a 

 very new or unusual circuit (from the standpoint of the speech distor- 

 tion) with one of common experience, several successive tests are 

 made upon the new circuit until no further increase in articulation 

 with practice appears. When it is impossible to intersperse the tests, 

 the data may be corrected to a given state of practice by means of 

 curves which were obtained in the following way. Although as will 

 be seen, this procedure is valid only under certain restrictions, which 

 will be discussed, such a correction will always tend to correct the 

 data to a more comparable basis. 



In Fig. 4-a a practice curve is shown that was obtained for a crew, 

 from two series of tests that were separated by an interval of three 

 months. The dots represent tests that were made upon a circuit 

 which uniformly transmitted a frequency range from 100 to 5,500 

 cycles. The circles were obtained from a circuit of the type shown 

 in Fig. 7 involving the carbon transmitter. In both cases the various 

 articulation values correspond to different received speech levels. 

 The crosses represent similar results that were obtained with a different 

 crew on the latter type of circuit. 



In Fig. 4-c the data of the first three tests in Fig. 3 are shown. 

 In this case the distortion was varied and the received speech level 

 held constant. As previously stated, in so far as was known, the 

 crew had no previous experience with these types of speech distortion 

 so that the practice for the various types of distortion ought to be 

 comparable. 



All of the solid curves are graphs of the following equation 



(1 -s') =^ (1 -sr-, (1) 



where S' — decimal value of syllable articulation obtained on a given 

 circuit at one stage of a crew's career, 

 5 = the value obtained on the same circuit at a later stage of 



the crew's career, 

 X = a number called the practice factor. 



The values of the practice factor x that were necessary in order to 

 fit the observed values are shown in the figure. It is impossible to 

 state definitely that a crew has uniform practice with various types 

 of distortion for the reason that experience is cumulative. A crew's 

 experience with one type of distortion may be of aid in the under- 



